In this tutorial, we will demonstrate how to measure the Ki67 index in breast cancer tissue using QuPath. The Ki67 index is a crucial marker for evaluating cell proliferation, providing insights into tumor aggressiveness and aiding in treatment decisions.
We will be working with a Ki67-stained whole-slide image (WSI) from the ACROBAT dataset. For demonstration purposes, we’ll use the image “0_KI67_val.tif” from the validation dataset “valid.zip”. You can import this image by following the steps outlined in our first article, “Introduction to QuPath.”
Our demonstration image uses H-DAB staining, a technique explained in our previous article. Here’s a brief recap:
In Ki67 staining, DAB highlights Ki67-positive nuclei with a brown color, indicating actively proliferating cells. The intensity and distribution of the brown stain provide insights into the tumor cells’ proliferation rate.
The Ki67 index is widely used to measure tumor cell proliferation. In H-DAB-stained tissue, brown-stained nuclei represent Ki67-positive cells, while blue or purple nuclei represent non-proliferating cells. Before detecting and counting cells, it’s essential to identify the correct regions of interest (ROIs) and understand what to include in the analysis.
To accurately measure the Ki67 index, we must later focus on regions with identifiable tumor cells. KI67 index is not measured within the whole sample, but regions with clearly identifiable tumor cells, excluding non-tumor components like stroma or necrosis. Areas where Ki67 expression appears most prominent, often called “hotspots,” are ideal for counting. Avoid regions with clumped or indistinct cells, as this can lead to inaccurate results when using QuPath’s automated tools.
The standard approach involves counting at least 500 to 1,000 tumor cells in areas of high magnification, focusing on those with the highest proliferation. The Ki67 index is calculated as the percentage of Ki67-positive nuclei out of the total number of nuclei in the tumor region, using the formula:
For this tutorial:
In the previous article “Tissue Segmentation Using a Pixel Classifier in QuPath” we covered how to differentiate between various tissue regions. As you proceed with this tutorial, ensure that you have already completed the tissue segmentation step.
Depending on your tissue sample, there are different approaches for moving forward. You can choose to apply the following steps across the entire tumor/epithelial area previously classified by the pixel classifier. In this case, both cell detection and Ki67-positive cell detection will be performed for the entire region. While this area may not represent a specific hotspot, you can later draw annotations to focus on specific subregions or hotspots. QuPath will automatically display the number of cells and Ki67-positive cells for each annotation you create within the broader tumor/epithelial area. Alternatively, if your tissue is large or a distinct hotspot is visually evident, focusing on that specific area can produce faster and potentially more accurate results, especially in terms of classification parameters like threshold adjustments. For this tutorial, we will begin by analyzing the entire tissue section.
Before we can specifically count Ki67-positive cells, we first need to detect all cells in the selected region.
QuPath detected a total of 36362 cells within my tumor/epithelial region.
Now that general cell detection is complete, we can focus on detecting Ki67-positive nuclei to measure the Ki67 index, which is the percentage of Ki67-positive cells relative to the total cell count. Continue working with the previously annotated area where cell detection was completed.
In this case, QuPath detected 5.37 % Ki67-positive cells out of the previously detected total of 36362 cells within the tumor/epithelial region.
After completing both cell detection and positive cell detection across the entire tumor/epithelial region, you can now focus on identifying a hotspot to calculate the Ki67 index more precisely. While QuPath does not automatically highlight hotspots, you can visually select an area that appears to have the highest concentration of Ki67-positive cells. Hotspots are typically identified either through manual inspection or semi-automated methods like heatmaps.
It’s important to note that this annotation should be placed within the larger annotation of the tumor/epithelial tissue. Once the annotation is created, QuPath will automatically calculate and display the number of detected cells along with the percentage of Ki67-positive cells within the selected region.
In this example, QuPath detected 1807 cells within my annotated hotspot, with 14.1% classified as Ki67-positive.
The final step involves interpreting the Ki67 index. Generally, a higher Ki67 index indicates a more aggressive tumor.
This tutorial is intended for learning and technical understanding of KI67 analysis in QuPath. While QuPath and similar tools are increasingly used in clinical diagnostics, manual review by a trained pathologist is obligatory for accurate interpretation.
Please note that the images used in this article have not been reviewed by a pathologist. As such, the percentage of Ki67-positive cells may not be pathologically accurate. This tutorial is meant to demonstrate the general principles of Ki67 scoring in QuPath and is not intended for clinical or diagnostic use.